Performance Analysis for Image Size Conversion Using Lapped Transforms

The size conversion method for images using lapped transforms provides superior performance than the conventional resizing method in the DCT domain. In this paper, the theoretical analysis is given for the resizing method in the lapped transform domain. It is shown that the resizing method in the lapped transform domain provides much better impulse responses and frequency responses than the conventional methods in the DCT domain. Computer simulations show that the conversion method in the lapped transform domain significantly improves the peak signal to noise ratio (PSNR) results. Specifically, the proposed method reduces the blocking effect to very low level in the conversion process, if it is used to convert the images which are compressed at low bit rates